Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads

[Tigerick]

Wildcat Lake (WCL) Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing Raptor Lake-U. WCL consists of 2 tiles: compute tile and PCD tile. It is true single die consists of CPU, GPU and NPU that is fabbed by 18-A process. Last time I checked, PCD tile is fabbed by TSMC N6 process. They are connected through UCIe, not D2D; a first from Intel. Expecting launching in Q1 2026.

		Intel Raptor Lake U	Intel Wildcat Lake 15W?	Intel Lunar Lake		Intel Panther Lake 4+0+4
Launch Date		Q1-2024	Q2-2026	Q3-2024		Q1-2026
Model		Intel 150U	Intel Core 7	Core Ultra 7 268V		Core Ultra 7 365
Dies		2	2	2		3
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		Intel 18-A + Intel 3 + TSMC N6
CPU		2 P-core + 8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		4 P-core + 4 LP E-cores
Threads		12	6	8		8
Max Clock		5.4 GHz	?	5 GHz		4.8 GHz
L3 Cache		12 MB		12 MB		12 MB
TDP		15 - 55 W	15 W ?	17 - 37 W		25 - 55 W
Memory		128-bit LPDDR5-5200	64-bit LPDDR5	128-bit LPDDR5x-8533		128-bit LPDDR5x-7467
Size		96 GB		32 GB		128 GB
Bandwidth				136 GB/s
GPU		Intel Graphics	Intel Graphics	Arc 140V		Intel Graphics
RT		No	No	YES		YES
EU / Xe		96 EU	2 Xe	8 Xe		4 Xe
Max Clock		1.3 GHz	?	2 GHz		2.5 GHz
NPU		GNA 3.0	18 TOPS	48 TOPS		49 TOPS

As Hot Chips 34 starting this week, Intel will unveil technical information of upcoming Meteor Lake (MTL) and Arrow Lake (ARL), new generation platform after Raptor Lake. Both MTL and ARL represent new direction which Intel will move to multiple chiplets and combine as one SoC platform.

MTL also represents new compute tile that based on Intel 4 process which is based on EUV lithography, a first from Intel. Intel expects to ship MTL mobile SoC in 2023.

ARL will come after MTL so Intel should be shipping it in 2024, that is what Intel roadmap is telling us. ARL compute tile will be manufactured by Intel 20A process, a first from Intel to use GAA transistors called RibbonFET.

 

[poke01]

I mean 8. There's a video where it compares all those generations. 8.1, 10 and 11. It's progressively more bloated, more memory use and such

This is also true for the macOS and Linux kernel they get larger with more features being added. Difference is Microsoft doesn’t give a crap about Windows optimisation now.

Same process, same efficiency. GPUs are embarassingly parallel so much easier to catch up, hence why Apple doesn't get decisive advantage in GPU vs competitors.

Apple had to overhaul its M5 GPU architecture to match the B390 at similar wattage in Cyberpunk. The GPU in the M4 is much weaker despite it being on N3E.

Then there is the other side of GPUs, compute. Which the B390 is firmly behind the M5.
It’s not as simple as using the same process so you get similar efficiency. Design and optimisation ultimately matters.

 

[511]

Then there is the other side of GPUs, compute. Which the B390 is firmly behind the M5.

which GPU RT in Blender Yes but what else it's winning in compute ? no one has done GPU benchmark testing like FP32/FP64 those are the traditional compute benchmark

 

[Joe NYC]

You're suggesting no iteration, but a straight shrink. Waste of everyone's time.
Medusa actually has new fabric/CPU/yaddayadda IP.

Shrink would have been the minimum. Lack of LP cores a big gap for AMD in mobile.

It's late'26 EEP; actual volume is 2027.

Which is why it is a problem. And I am not suggesting the entire line up. Just the Kracken shrink would have been enough to stay relevant.

PTL-H has the same-ish 1t perf as the antique RPL-H from 2023.

RPL was quite literally the biggest boon Intel had; cheap and nice volume drivers.
Every part after RPL is horribly margin dilutive until *maybe* NVL-U 4+0+4.

We will see how things go with Intel price increase (and reduced RPL supply). AMD may be a short term winner, due to shortages (H1 2026) but longer term (H2 2026) PTL looks good vs. somewhat dated AMD design now on a trailing node.

It will be hard for AMD to gain share in mobile in H2 2026 with a stale line up.

Gorgon point is kind of an insult to people's intelligence.

 

[adroc_thurston]

Shrink would have been the minimum

No, it's useless.

Lack of LP cores a big gap for AMD in mobile.

Very small problem.
They just don't care about mobile, RMB was like the last time they did (4 years ago).

Which is why it is a problem. And I am not suggesting the entire line up. Just the Kracken shrink would have been enough to stay relevant.

They're relevant either way.

We will see how things go with Intel price increase (and reduced RPL supply). AMD may be a short term winner, due to shortages (H1 2026) but longer term (H2 2026) PTL looks good vs. somewhat dated AMD design now on a trailing node.

who cares man, H2'26 is Helios ramp.
and Venice ramp, just a double-whammy of stuff to be busy with.

It will be hard for AMD to gain share in mobile in H2 2026 with a stale line up.

They don't care. Lisa cares even less.

Gorgon point is kind of an insult to people's intelligence.

It's a lineup of de-facto ASP hikes which is funny.
But normal, TSM is hiking wafer ASPs up too.
It's actually a problem for NVL/RZL too but that's a later story.

 

[clemsyn]

Just curious..Why does the IGPU of Panther Lake still built by TSMC 3nm? Is it just because they have a contract or does it show Intel having issues to build the IGPU in their fabs?

 

[511]

Just curious..Why does the IGPU of Panther Lake still built by TSMC 3nm? Is it just because they have a contract or does it show Intel having issues to build the IGPU in their fabs?

both

 

[Joe NYC]

Ticks are dead because modern shrink is like 15% speed 30% power best case.

30% power savings sounds quite fantastic.

 

[Khato]

I'd go with Intel wanting to use what 18A capacity they have available for the main SoC tile rather than the 12Xe GPU tile. The 4Xe GPU tile is on Intel 3.

If anything Intel using multiple different TSMC and Intel processes on the same Xe3 IP is indicative of Intel Foundry finally getting a bit better with standard design flows instead of Intel specific.

 

[adroc_thurston]

30% power savings sounds quite fantastic.

Active power; contributes snot to BL.
Also irrelevant, you want the shrink for speed.
15% speed on Zen5 is snooze. Zen6 it is.

 

[Joe NYC]

Very small problem.
They just don't care about mobile, RMB was like the last time they did (4 years ago).

A combination of DDR5 prices and OEMs not giving a damn about AMD kind of doomed it.

who cares man, H2'26 is Helios ramp.
and Venice ramp, just a double-whammy of stuff to be busy with.

You mean the people responsible for mobile SoC development are just standing there, chewing gum and looking at guys doing Venice, Mi455 and Helios?

That what it seems like.

 

[dullard]

Just curious..Why does the IGPU of Panther Lake still built by TSMC 3nm? Is it just because they have a contract or does it show Intel having issues to build the IGPU in their fabs?

Intel does not have the equipment or the substrates (wafers) to do it. They are limited on both fabs and on supplies right now.

Look how little advanced fab capacity Intel has in 2026.

https://www.techpowerup.com/319704/intel-10a-1-nm-class-node-to-enter-mass-production-in-2027#g319704-1

AI and data centers have taken much of the raw materials.

 

[Joe NYC]

They don't care. Lisa cares even less.

Funny, how Jensen, who has a much bigger fish to fry, still cares about client mobile space, where NVidia does not even exist, while Lisa does not care.

Which is your interpretation - that Lisa does not care.

The reality is that AMD got caught with their pants down and will be scrambling to get back into a competitive position. That's what a smart CEO would do.

 

[dullard]

30% power savings sounds quite fantastic.

Don't argue with him, he knows absolutely nothing and can't back up a single post with links.

 

[Joe NYC]

Don't argue with him, he knows absolutely nothing and can't back up a single post with links.

I kind of disagree. What you can gather from "links", I already mostly know. It matters how you interpret things.

 

[adroc_thurston]

A combination of DDR5 prices and OEMs not giving a damn about AMD kind of doomed it.

Not my point.
RMB was the last time AMD iterated mobile uncore.

You mean the people responsible for mobile SoC development are just standing there, chewing gum and looking at guys doing Venice, Mi455 and Helios?

Yup.

Funny, how Jensen, who has a much bigger fish to fry, still cares about client mobile space, where NVidia does not even exist, while Lisa does not care.

He owns the place.

Which is your interpretation - that Lisa does not care.

The reality is that AMD got caught with their pants down and will be scrambling to get back into a competitive position. That's what a smart CEO would do.

Lisa does not give an ounce of shit about client stuff.
It's a poverty market with margins being actively thinned by a bunch of factors.

They're not scrambling anything nor adjusting their roadmap in any way.
It's Medusa (2027) > Grimlock (2029) > Zen8 mobilestuffs (2031).

Also none of this is Intel-related.
goto Zen6 thread for that and tag me.

 

[LightningZ71]

It makes perfect sense why gaming mobile PTL would be 2+8+4. Lighting up 4 P cores to get more performance than 2 + 2 e cores would blow the power budget enough to leave nothing for the rest of the chip in a 15-20 watt envelope. Any game that absolutely requires more than 2 P cores at above E core performance to perform well is either not going to work well with an iGPU or can't do well in a handheld power budget to begin with. Pushing 30 watts to make it make any sense raises your build cost for a larger battery and beefier cooling and gets beat by Strix Halo.

 

[OneEng2]

Lisa really needs to re-think it if she wants AMD to capture 40% of client market. It is not going to happen if AMD releases new products only every 2.5 years.

Intel really needs to re-think if they ever want to turn a profit again. They have ceded most of the high profit markets to AMD with their current strategy.

AMD is going to be very strong in desktop, but mobile is a bigger market, and AMD is not taking it seriously enough.

I would agree; however, which one makes the most profit? Intel seems to be ignoring DC which I see as a much bigger crime (from a profit standpoint).

Additionally, mobile is a huge volume. Selling 18A chips at commodity pricing for Intel (in mobile) seems suicidal to me.

You don't think battery life matters to users? I think most people would happily take marginally worse page load times for massively better battery life.

Agree. I think that battery life is likely the #1 selling point in mobile right behind price of course .

It is a big market. AMD is not going to get the premium share if it is not strong enough in the mainstream segment.

The premium share always goes to the "best" (whatever that means) in the market. In mobile, I think this is quite a complex question. In DC .... it's pretty simple in comparison.

Assuming Intel can pull another 6-8% IPC out of Nova Lake and AMD gets 12-14% out of Zen 6 it will come to frequency for the ST crown.

I worry that traditional downfalls of BSPDN will cripple Intel's ability to scale clock speed. Ironically, the DC is where I think 18A would shine the brightest .... if Intel had a chip worth selling in that market.

 

[Josh128]

Assuming Intel can pull another 6-8% IPC out of Nova Lake and AMD gets 12-14% out of Zen 6 it will come to frequency for the ST crown.

AMD will get ~10-12%, not 12-14%.

 

[adroc_thurston]

AMD will get ~10%, not 12-14%.

They'll get low teens (think 13%) as usual.

 

[Josh128]

They'll get low teens (think 13%) as usual.

Thats better than their own projections, but OK.

 

[adroc_thurston]

Thats better than their own projections, but OK.

Their internal projection for Venice was 1.5x skt score and we're waaaay past that mang.

 

[511]

Their internal projection for Venice was 1.5x skt score and we're waaaay past that mang.

SKT?

 

[adroc_thurston]

SKT?

socket.

 

[OneEng2]

AMD will get ~10-12%, not 12-14%.

I wonder how well the performance overall will be for Intel vs AMD though? If Intel has issues clocking the design up (and retaining yields), then AMD's performance will likely eclipse Intel in the next gen lineup.

It is even worse if you look at the margins that are made at higher binning vs lower binning. Again, I worry more about Intel's ability to make money than them producing chips that have higher IPC than AMD's next gen design (which I expect is a very distinct possibility based on the data currently available).

 

[Josh128]

Their internal projection for Venice was 1.5x skt score and we're waaaay past that mang.

​

skt scr ≠ IPC. AMD pj +10-15% IPC fr Zen 5, 10% fr Zen 6. 2 yr old trufe bmb dawg.